Why are reforms non-sustained?: a case study of Turkey

(1)

I .« S '» ■■.*■> í Éí’

ЩУ¥· I p “ ^ r p ñ p f j f ; ІІ^Щ «

_{iî’i i î ï} _{Л.|{.іа l a L s t j i l t s î s s i a t ^ . ^ · >g>%fзд g Л*Ьіа«4^г 5 « л 's lí-W}

*. â fíi^ü

^ I; S it. li; î^ff :^ΰ· * -'J Ï м‘ ,ч it щ .- х “'.;.Îİ4 Â л ir · lan-, ·?ΙΜΙ·. y f Μ (HH«t‘ ♦ ' *■>■■■' u»b" !"■ . .*■ ; ’ ■ ? '· -î-· «'■' ■ Λ 4. . » · «мГ «- ■ α, ·· ии.^ Λ· іийЧ -íW»*· 1»J . J.·, Гу?"“ •^’■f · ■ ^‘* - ■ ' Î лГ*·*:. •i'Mİmif' îrtiU ш4і/ <r«J» » ^ЖІПІИ'» ' 'И*'*' '■Ни'* ^ «wA···* і ,йі»

·♦■ . , ‘ «к* ' ■. , ' .·" ♦ ». ’ ·■ . ■ '· ·'· -f··'·^' « <«#<■ Я> ,ІЧ«»»Ч*«4». « f л МЯП..М )*. *Ч '»#· Ч.

(2)

WHY ARE REFORMS NON-SUSTAINED?; A CASE

STUDY OF TURKEY

A THESIS PRESENTED BY SaLAH JeLASSI TO

THE INSTITUTE OF

E

conomics and

S

ocial

S

ciences

IN

partial fulfillment of the

REQUIREMENTS

FOR THE DEGREE OF MASTER OF ECONOMICS

BILKENT UNIVERSITY

(3)

H J

1 ß ß i

(4)

1 certify that I licive read this thesis and that in my opinion it is fully adequate, in scope and in quality, as a thesis for the degree of Master of Science.

Assist. Prof. Izak Atiyas (Advi.sor)

1 certify that I have read this thesis and that in my opinion it is fully adequate, in scope and in quality, as a thesis for the degree of Master of Science.

I certify that 1 have read this thesis and that in my opinion it is fully adequate, in scope and in quality, as a thesis for the d e g i^ of Master of Science.

t l j ^

Assist./Prof. Syed F. Mahmud

Aj)proved by the Institute of E c o n o n ^ and Social Sciences. Director:

(5)

ABSTRACT

W H Y A R E R E F O R M S N O N -S U S T A IN E D ?: A C A S E S T U D Y OF T U R K E Y

Salah Jelassi

M A S T E R O F E C O N O M IC S Supervisor: Assist. Prof. Izak Atiyas

September, 1996

The factors affecting the reform sustenance are the main topic of this thesis. We focus on the tax financing package that we assume necessary for a large-scale reform progi'cim. We show theoretically how direct cincl indirect taxes may lead to the actual reform duration cind costs be under estimated by the labour class. Factors such as tax-shifting and access to financial adaptation were deterministic in our claims. Using Turkish time series data for public wages and salaries, we try to show tax-shifting empirically by estimating a real wage determination equation. Partial evidence for tax-shifting in real public salaries was found. A simple recapitulative model establishing the link with the results derived by the related literature wcis postulated.

Keywords: Reform, Sustenance, Tax-Shifting, Financial Adaptation, Wage Determination Equcition.

(6)

ÖZET

T Ü R K İ Y E ’D E R E F O R M L A R N İÇ İN D E V A M L I D E Ğ İL ?

Salah Jelassi

Yüksek Lisans Tezi, iktisat Bölümü Tez Yöneticisi: Yrd. Doç. İzak Atiyas

Eylül, 1996

Reformların devamlılığını etkileyen faktörler bu tezin temel konusunu oluşturmaktadır. Geniş kapsamlı bir reform programı için gerekli olduğnu varsaydığmız vergi finansman paketi üzerinde yoğunlaşmaktayız. Direk ve dolaylı vergilerinin gerçek reform süre ve maliyetlerinin işçi sınıfı tarafından daha az tahmin edilmesine sebep olduğnu teorik olarak göstermekteyiz. Vergi zıplamaları ve finansal adaptasyonlar çalışmamızda deterministle olarak ele alınmıştır. Vergi zıplamasını ampirik olarak gösterebilmek için reel ücreti belirleyen denklem tahminini Türk kamu sektörü ücret datasını kullanarak gerçekleştirdik. Reel kamu ücretlerinde vergi zıplaması olduğna dair bazı sonuçlara ulaştık. Basit özet bir model ilgili literatürle bağicintıları göstermektedir.

Anahtar Kelimeler: Rerform, Devamlılık, Vergi Zıplaması, Finansal Adaptasyon lar, Ücret Belirleme Denklemi.

(7)

ACKNOWLEDGEMENT

I would like to express my gratitude to Dr. Izak Atiyas for his valuable guiding throughout this work. Special thanks go to Dr. Osman Zaim for his contributions and comments and for having read cind evaluated this thesis. I also would like to thank Dr. Syed F. Mahmud for his evaluation and for his continuous support throughout my career. I appreciate the help of my friend and home-mate Nafaa and his patience in writing this work.

I would like to thank all the members of my large family and especially my sister Zohra for her never-ending moral and financial support. Finally, I dedicate this work to my ’’ dearest” Soheir without whom it would have been much harder to accomplish.

(8)

Chapter 1 Introduction: Literature Review

'I'he main aim of political-economics is to view economic phenomena that seem to be irrational, as a rational outcome of the actions of self-interested individuals nuiximizing their utility.

In Tui'key, an interesting feature produced by the 1994 and 1996 elections, is the increasing votes going to the Islamic fundamentalist party which is viewed to be against the reform process that was started in the eaidy 1980’s.

This question falls in the area of the so called new political economy, which tries to view apparently irrational outcomes as an interaction between political and institutional factors and macroeconomic policy aspects using the tools of mainstream economic analysis and game theory.

As a rejection of the classical idea of a benevolent social planner and exogenous policy determination, researchers in this area have developed theoretical models where economic policy is determined endogenously as an interaction among different agents each challenged by his self-interest.

A politically sustainable reform package can be defined as one for which there is sufficient political support to carry out the reforms to their completion. This political support can take various forms; for example the re-election of the government that started the reforms, a sufficient degree of parliamentary support for reforms, and a large degree of socicd acceptability. In contrast, frequent labor union organized strikes, an increasing degree of political polarization and ci slow-acting parliament

(11)

are considered as signals of reform failure. The related literature has been tackling this iDi’oblem from both a normative and a positive perspective. The normative perspective is particularly present in the analysis of how institutions affect policy Ibrmcition. This perspective is very relevant due to the critical role of institutions in determining the reform-environment. The status of the central bank, the social choice process, the state of the bureaucracy and the strength of political parties and their discipline are just examples of such critical institutions. The positive perspective is cilso important from a policy point of view because it can give insights on the type of policies that cire more likely to emerge from specific political cind institutional settings, and therefore assist in the design of politically sustainable policies.

In what remains in this introductive chapter we will provide a literature survey of what kind of structural features and economic policy choices and how they interact to affect the reform process.

1.1 structural Features:

As pointed earlier the feasibility and political sustainability of an economic reform is affected to a large extent by the underlying structural features. In this context, we examine how the pre-reform economic environment, the institutional structure and the nature of the politiccil system come out to be deterministic factors in the reform success.

1.1.1 Pre-reform Economic Environment:

CHAPTER 1. INTRODUCTION: LITERATURE REVIEW

2

A wide number of pre-reform environment conditions have an impact on macroe conomic policy reform. Rodrik (1993) argues that macroeconomic imbalances determine the urgency of a reform. In circumstances of extreme macroeconomic imbalances, a reform can benefit from the unsustainability of the status c[uo to gain the necessary support for its adoption. Laban and Sturzenegger (1994) carry this point further and show theoretically that even if there is no uncertainty about the evolution path of the economy of getting worse, it may be optimal to posti)one the

(12)

reform, this stems from the fact that stabilization is costly and involves risk. The cuialogy is made with a patient who needs to undergo a risky operation. The high risk of not surviving may make it optimal to postpone the operation and live another year with certainty although the patient knows that he must eventually be operated on and his condition is deteriorating.

Moreover, a country with a higher per capita income is more prepcired to withstand the adjustment costs that are associated with the reform. However, this better initial .situation may form an obstacle for the reform to pass due to the fact that tluj majority may stick to the status quo faced with the reform costs and its uncertain benefits. This leads to the fact that reforms in countries with a more unequal income distribution need to address the impact of these measures on lower- income groups that are less likely to stick to their initial situation.

On the other hand, a skewed income distribution can also form an obstacle to an economic I'eforrn as it is often associcited with a sharper degree of polarization of the society that makes it more difficult to reach consensus on certain policy issues.

Microeconomic distortions affect the burden-sharing of the reform costs across population groups together with the amount of resource reallocation associated with the removal of the distortions.

Credibility is another factor that can be deterministic in a stabilization program’s failure. In a country with a history of failed reform attempts, the behavior of the public will be inconsistent with the attainment of reform goals, thus effectively dooming the reform plan to failure (Calvo (1988)).

Fernandez and Rodrik (1991) show that in the presence of identity-uncertainty, that is when individual gainers and losers can not be identified ex-ante, there will be a bias towards the status quo. This may take place even if there is no aggregate- uncertainty regcirding the consequences of the reform. They show that a reform may fail to be adopted although it may benefit expost. Conversly, a reform may be adopted originally but fails to gain the necessary support to sustain it. The writers apply their cinalysis to tiuide liberalization as an interesting example of the delay of a. widely cigreed-upon beneficial reform. By assuming that in order lor a worker to sliift from one sector to another, he should incur a two component cost: A general investment cost incurred prior to switching sectors and cui individual specific cost

(13)

element incurred only upon actually switching sectors and is revealed only when the general investment cost is incurred; the previously claimed results of beneficial- reform delay is derived.

4

1.1.2 Institutional Structure:

The second structurcil feature that heavily influences the economic reform process is the underlying institutional structure. The examples of such an interaction are numerous.

Liberalization of foreign direct investment is unlikely to be effective in the absence of a well defined system of property rights or in the absence of a functional banking system. Fiscal consolidation may be more difficult to attain in an institutional setting with decentralized expenditure centers. The speed of labor reallocation across sectors following structural reforms depends, among other things, on the flexibility of labor market institutions.

Another important and quite stressed institutional factor is the status of the central bank. The autonomy of the centred in choosing the instruments of monetary policy is described by: (i) the influence of the government in determining how much to borrow from the central bank, and (ii) the nature of the monetary instruments under the control of the central bank. If the government can influence the quantity and the conditions on which it borrows from the central bank, it also influences the creation of monetary base and lessens the economic independence of the central bank. Grilli, Masciandaro and Tcibellini (1991) claim that one of the alleged benefits of an independent central bank is that it is more credible in its resolve not to engage in public financing. This may affect the treasury’s behavior, reducing the likelihood of unsustainable debt policies. They show empirically that central bank independence is on average associated with lower inflation, irrespective of political institutions and budgetary problems.

On the other hand, constraints on fiscal policy imposed by the legislature, such as balanced budget requirements, although may reduce policy flexibility, if adhered to, will enhance fiscal discipline. Bayounii and Eichengreen (1995) provide evidence indicating that U.S states with stringent statutory and constitutional restrictions

(14)

on deficit spending and debt issue have more stable fiscal balances.

Finally, we should note that the institutional structure is a complex factor affecting reform sustainability stemming from the fact that it is not invariant. Rather it tends to change during the reform process with a velocity that depends on country- specific characteristics.

5

1.1.3 Nature of the Political System:

The third structural feature that is related to the reform process is the nature of the political system. It simply determines the environment in which the reform will survive.

Roubini and Sachs (1989) stress in their analysis the fact that governments are not monolithic entities of standard economic models that have full control of the policy instruments and that manage them according to a stable and a well- defined objective function. When power is dispersed, either across branches of the government (as in the U.S), or across many political parties in a coalition government (as is typical in Turkey), or across parties through the alteration of political control over time, the likelihood of inter temporally inefficient policy is heightened. The budget problem for a coalition government arises from three factors. First, the individual coalition partners in multi-party governments have distinctive interests and distinctive constituencies. There is no single uniform objective function for the Vcirious political parties in the government. There is likely to be a fundamental prisoner’s dilemma with respect to budget cuts: all the partners of the cocilition may prefer comprehensive budget cuts to a continuation of large deficits, but each cocilition partner may have the incentive to protect its particular part of the budget against the austerity measures in the absence of strong coordination between members of the coalition to produce the ’’ cooperative” outcome, the noncooperative solution of no-budget cutting is quite likely to arise. Second, individual coalition partners will often have enormous power to prevent a change in the status quo, though they will not typically have the power by themselves to implement a positive program of change. In other words, coalition members will have a veto against change. Even a very small party in a multi-party coalition can have enormous power by virtue of its ability to brecik up the government. Moreover, a coalition will

(15)

typically divide responsibilities over various parts of the budget among the various members (this is certainly true in part by virtue of the distribution of ministerial positions among the parties). Third, the enforcement mechanisms among coalition partners to assure the cooperative outcome will often be very weak.

The importance of the time-horizon of the government for budget policy was stressed by Alesina and Tabellini (1987). They emphasize thcit when political power alternates randomly between competing political parties, each government will be tempted to leave a legacy of high debt for its successor whose spending priorities it is not likely to share. The high debt restrains the spending by the next government, but the current government cares little about the next government’s spending in any case. Presumebly, the more rapid the turnover of the government, the more important would be this deficit bias effect.

Alesina and Perotti (1993) using assembled data for seventy countries show that politiccil polarization that often results from unequal income distribution increases socio-political instability. Income inequality increases social discontent and fuels social unrest. The latter, by increasing the probabilities of coups, revolutions, mass violence or more generally by increasing policy uncertainty and threatening property rights, has a negative effect on investment and as a consequence reduces growth. The link between income inequality and growth is viewed through another channel by Alesina and Rodrik (1991,1993). They claim that in more unequal societies, the demand for fiscal distribution financed by distortionary taxation is higher, causing a lower rate of growth.

In the case where voters are perceived to be ’’ myopic” , the phenomenon of what we call ’’ election-politics” may arise. Incumbants would think twice before cidopting reform measures that may cause short-term costs for fear that this would lead to electoral defeat. Alesina and Roubini (1992) provide some, evidence that the policy stance of governments change significantly in electoral years. Milesi-Ferretti (1992) constructs a simple model in which it may be rational for a government facing an election not to embark on a disinflationary program, even though voters’ behavior is forward looking. This may be due to three reasons. First, designing an efFicient disinflationary program is costly: by devoting resources to this purpose, the government forgoes ’’ popularity” or the possibility to maintain current levels of

(16)

government consumption. Second, disinflation entails unavoidable output costs, but these costs are uncertain. Third, the government cares about being in power, for reasons that go beyond the possibility to implement its most preferred policies.

1.2 Economic Policy Choices:

Another body of the related literature has been emphasizing how the economic policy choices affect the reform success.

CHAPTER 1. INTRODUCTION: LITERATURE REVIEW 7

1.2.1 Speed, Depth and Sequencing of reform:

In our previous discussion our arguments were valid for any type of a reform. Howerver, here we differentiate between macroeconomic stabilization such as reduction of inflation or of fiscal imbalances; and economic reform such as trade liberalization, the removal of price controls, privatization and restructuring. The difference is that stabilization policies are typically short-term while some economic reforms take a longer time to be implemented.

Gradualism versus ” big-bang” type of a reform is a genuine question to cirise in the case of a structural large-scale transformation. On the other hand, in the case where reform packages include several measures, the appropriate sequence in which those measures are to be adopted is another matter that is open for debate.

Dewatripont and Roland (1992) using a simplified model of restructuring where a reform-minded government faces a set of sectors with a work force which is heterogeneous in terms of relative opportunities outside their current jobs, shows that gradualism can be better than full immediate reform when the financial cost of compensations necessary to respect the existing political constraints are very high compared to the allocative gain of immediate restructuring. Gradualism is less costly for the budget because it allows for better rent extraction from some groups of workers.

In contrast, Mcirtinelli and Tommasi (1994) claim that by widening the scope of efficiency improving reforms, the government is more likely to gain the support ol larger segments of the population, particularly il the losers ol each particular measure

(17)

cire benefited by other measures. If the government needs to pass a threshold of popular support at each step, a gradual process risks being stopped at each stage by the group being hurt at that point. Note that this provides an explanation for why Latin American countries that failed when implementing gradualist programs in the ecirly 1980s, have been relatively successful in later undertciking more comprehensive attempts at reform.

CHAPTER 1. INTRODUCTION: LITERATURE REVIEW 8

1.2.2 Distributional Aspects and Foreign Assistance:

Distributional aspects play a role in economic reform since the economic impact of these usually Vciries across population groups. Measures that soften the impact of the reform costs on the low income groups such as social safety nets, may enhance the political sustainability of reforms.

Mussa (1984) argues that it is not always the case that those more directly affected by reforms are the lowest-income groups. In such cases, redistributive measures intended to protect the reform may involve compensation to special interest groups that benefit from the pre-reform policies. Moreover, some stabilization measures, such as inflation reduction, may have positive distributional effects. Various studies suggest that inflation is a regressive tax whose burden falls disproportionately on lower income groups with limited access to hedging instruments (Cardoso(1992)).

Redistributive schemes that may seem to be necessary for the survival of a reform may represent a problem of implementation. From one side, a weak institutional structure and a lack of technical competence of the civil service may complicate such a process. On the other hand, for poor countries with limited access to capital markets, and countries facing adverse external conditions such as a war or a drastic reduction of transfers from abroad, budgetary problems may constrciin the feasibility of such schemes. Such budgetary problems may be lessened by foreign assistance. In general, foreign credit and assistance allows a country to smooth consumption over time when faced with short-run costs ol stabilization and reform. This widens the policy options of a reforming government, and may enable a country to start reforms earlier or choose a quicker pace of reform.

(18)

1.3 The Interplay between Structural Features

and Economic Policy Choices:

The structural institutional and political features and economic policy choices interact with each other to determine the reform ’’ destiny” . The examples of such an intei'ciction ci.re numerous some of which will be discussed in a comparatively larger detail in this section.

1.3.1 Initial Macoeconomic Disequilibrium:

When initial macroeconomic disequilibria are extreme, the ’’ status quo” is clearly unsustainable. In such situations, it may be easier to gather political support to reform measures. Rodrik (1993) argues that the benefits from stabilization Ccin make it politically easier to simultaneously implement reforms that involve a greater amount of redistribution among population groups. This can result because the ’’ losers” from reforms are compensated with the gains from the stabilization.This suggests that the failure to combine reforms involving redistribution with measures that beneht the entire population may account for the difficulties in gaining enough sustenance for the reform.

Krueger (1993) argues that the suspension of politics in a period of crisis Ccin facilitate the implementation of reform measures. This is due to the fact that such measures can become self-sustaining over time, as new constituences supporting the reform emerge throughout the process.

9

1.3.2 Sequencing and Speed of Reforms:

Fernandez and Rodrik (1991) show that there are some cases where the reform is not implemented ex-ante although it may be optimal. Wei (1993), in the context of a trade liberalization, shows that adopting the reform grcidually may help over passing such difficulties. This stems from the fact that a gradual process of a reform may alter the nature of political system by building a constituency in favor of the continuation of reforms.

(19)

₁₀

Martinelli and Tommasi (1993) Show that time inconsistency problems may require the ’’ optimal” sequencing of reforms to be altered in order to guarantee its politiccil feasibility and sustainability. They assume that different reforms entail different winners and losers. Hence simultaneous implementation of all reforms although sometimes suboptimal, is politically sustainable.Once a reform has been adopted, its beneficiaries have no incentive to support further reforms that may hurt them and benefit other groups. The critical implicit assumption made by Martinelli and Tommasi is that no reversal of previous refoi'ms occurs if subsecpient ones fail to be adopted, so that ’’ early” winners have an incentive to withhold further support.

Conversely, Dewatripont and Roland (1993) assume reforms are rather comple mentary, in the sense that the interruption of the reform process implies a costly reversal of previously implemented ones. The reversal costs rises as the reform progresses. Consequently, a big-bang strategy is very costly to reverse which may yield its sustainability. By contrast, a gradualist strcitegy retains the option of reversed should reforms prove unsuccessful, and is therefore harder to sustain politically.

1.3.3 Speed of Reform in the Case of Restructuring:

Economic restructuring is a kind of an economic reform that involves sectors in which resources are used inefficiently, and in which realization of efficiency requires labor shedding and a more intensive use of resources. If the political situation is such that ’’ restructuring” requires the support of a majority of workers in the sector, these may need to be compensated to relinquish their jobs. Dewatripont and Roland(1992) argue that the task of the government to design workable compensation schemes is complicated by the fact of informational asymmetry in the sense that the government does not know how the workers differ in their employment opportunities outside the sector. This is a reflection of a more general problem of previously centrally-planned economies of the existence of ’’ hidden rents” from the current status quo that ¿rcross social groups such cis housing privileges, legal and illegal parallel activities..etc.

(20)

₁₁

1.3.4 Credibility and Political Uncertainty:

Another point mentioned earlier is how the lack of credibility and political uncertainty can lead to the failure of reforms. Failure of the government to fulfill its commitments in its previous reform attempts enhance such credibility problem and increase the degree of political uncertainty.

Rodrik (1989) suggests that the sustainability of the reform may benefit from an initial ’’ overshooting” , because this will enhance credibility by signaling to the public the seriousness of the government’s intentions. This would push the public to chase actions that take the government’s commitment to reform as established, thus facilitating reform.

Lack of credibility is likely to arise when the political system is fragmented or polarized. When the political system is fragmented, agents can come to doubt the government’s ability to ensure sufficient support for reform measures over the time, because several groups may have veto power. In this case, in which the political situation following the implementation of reforms is likely to become fragmented or unstable, with a status quo bias, an early and simultaneous implementation of reforms may be the strategy that ensures the political sustainability. If the economic reform reversal is not costless, the high costs associated with such tin act may preclude a rollback even in a difficult political situation. A more gradual approach may instead run into political difficulties, as the fragmented or polarized political system makes it harder to implement further reform metisures.

1.3.5 The Supply Response to Reforms:

The supply response to economic reforms is extremely important in determining their politiccil sustainability. When reforms fail to reach their expected results, or when the costs associated with the transition period cire higher than expected, reform reversal is more likely for several reasons. First, voters may inter from events that the implemented reforms are ineffective, in which case they will be against them. Second, voters may believe that the government has been ineffective in policy implementation, and therefore vote for opposition groups who are seen as being less favoralile to reform.

(21)

₁₂

A paper by Wijnbergen (1992) stresses the importance of a quick output supply response for the sustainability of reform. The author’s main argument is that opposition to price liberalization may be fueled by the fear of a low output supply response to freed prices. In such framework, a gradual removal of price controls can leiid to hoarding with the purpose of intertemporal speculation, as prices cire e.xpected to rise in the future. The fact that production is withheld from the market while inventories are being accumulated can generate shortages and therefore a perception of low output supply response. This perception can lead to the opposition to the removed of remaining controls, as individuals come to expect that higher prices will not be sufficiently compensated by an increased availability of goods. Conversely, an immediate full price liberalization can eliminate incentives for hoarding and intertemporal speculation, thus enhancing the response of sales to price decontrols. This enhanced response can strengthen the political support for price reform. Therefore, in this case immediate price liberalization is not only more efficient, but also more credible since political considerations are less likely to lead to future reversals.

1.3.6 Distributional Conflict:

The changes associated with the beginning of a process of economic reform will inevitably give rise to distributional conflicts. Alesina and Drazen (1991) use a model of ’’ war of attrition” to study this issue. The writers give three reasons that suggest modeling delayed stabilization as a war of attrition:

1

. In the political debate over stabilization, the distributional question of how the burden of higher taxes or expenditure cuts should be allocated is central.

2

. When stabilization occurs, it coincides with a political consolidation. Often one side becomes politically dominant. The burden of stabilization is sometimes quite unequal with the politically weaker groups baring a larger burden.

3

. Successful stabilizations are usually preceded by several failed attempts. Often the previous program appears to be similar to the successful one.

(22)

₁₃

For a war of attrition between heterogeneous individuals to give expected finite delay in concession under incomplete information, two features are important:

1

. There must be a cost to remaining in the fight, that is not conceding.

2

. The pciyoff to the winner must exceed that to the loser.

In their model, Alesina and Drazen assume that before stabilization the government is incurring a budget deficit with increasing debt which is financed through distortionary methods, mainly inflation taxes. A fraction

7

of the debt is fincinced through distortionary taxation. Now there are two interest groups in the economy and stabilization is achieved through the elevation of taxes sufficient to prevent further growth in the debt. An agreement to stabilize is an agreement on how these taxes are to be apportioned between the two groups. The ’’ loser” assumes a fraction a > 0.5 of the tax, and the winner a fraction

(1

— a). A value of a close to 1 represents a high degree of polarization or a lack of political cohesiveness. Taxes after stabilization are assumed to be non distortionary. Before stabilization each group knows his loss from distortionary taxation ^¿, as well as the distribution of his opponent’s losses F{0j).

A certain group will not concede as long as his expected utility from being subject to the distortion or that his opponent concedes thus paying only the fraction

(1

— a) of the tax, exceeds his utility from paying a fraction a and avoiding the distortionary losses Bi. The following main results are shown:

1

. The higher is

7

the earlier is stabilization.

2

. An increase in 9 will bring the stabilization earlier.

3. If a =

0.5

stabilization occurs immediately, and the closer a to

1

the later is the expected date of stabilization.

1.4 Conclusion:

Throughout this introductory chapter, we have seen that ensuring political support and sustenance of an economic reform is quite a complex matter that requires

(23)

14

a careful consideration of the political and institutional environment in which the reform is implemented, and of the distributive consequences they generate. Studies in this topic, clearly could not provide firm conclusions and a set of policy recommendations guaranteed to ensure the sustainability and success of reforms. However, the literature does provide useful insights and generalizations in the complex interciction between economic and political factors, whose Vcilidity and applicability requires a careful study of actual policy experiences in the reforming country cuid in others.

(24)

Chapter 2 Theoretical Foundations

2.1 Direct Taxes: W ho actually pays for them ?

Following any reform package that aims at trade liberalization, by removing price controls, implementing privatization and resti'ucturing, there should be ci financing tax package.

Our claim here is simple: Originiilly, an individual will take the decision to support a reform depending on the announced tax shares and his expected benefits from the reform. The point here, is that an individual may underestimate or, reciprocally, overestimate his costs due to the fact that the true incidence of the tax is not a matter that can be determined easily. Rather, the person who effectively pciys a tax is not necessarily the person upon who the tax is levied.

2.1.1 The Model:

In order to clarify our chiirn, we use Solow’s (1956) one-sector equilibrium growth model. We assume that the production side is governed by a constant return to scale technology:

‘ I would like to thank Dr. Osman Zaim for his potential contributions in this section.

(25)

CHAPTER 2. THEORETICAL FOUNDATIONS

₁₆

Y = F { K , L ) (

2

.

1

)

where: Y = aggregate output level, F = total capital,

L-- total labor force,

and F is a twice differentiable increasing concave function, homogeneous of degree one

Now, using the homogeneity assumption on F, equation (

4

.

1

) can be rewritten clS

r = i . f { k)

_(2.2)

where: k — ^ =the capital labor ratio.

and f " <

0

follow from the assumption described above.

Labor is assumed to be inelastically supplied as a fixed proportion of a population growing exponentially at a rate n over time v,:

L = Lo - U (2.3) or 6u = Lone"“ - nL that is where: L = ^_{a u} L L = ^ (2.4)

Capital is also assumed to be inelastically supplied, and its growth to depend on savings and depreciation. The gross saving out of output is assumed to depend on the capital-labor ratio. Description is assumed to have the simple form of a proportionate rate of capital. That is :

(26)

₁₇

K = s { k ) Y - ' ^ - K (2.5)

where

7

stands for the rate of depreciation.

Now, substituting (

2

.

2

) into (2.5) and rearranging yields the equation in (2.6):

k = s{k)L.f{k)--^.K k ^ s{k)L.f{k) K K k _ s{k)f{k) K k ■ 7 - 7 (2.6)

Now, we suppose that our economy is in the equilibrium pcith, so that both labor and capital cire fully employed. Thus differentiating (

2

.

1

) and rearranging yields the equation in (2.7);

8_2 _ 6F^6J^

^ 6 ^

8u ~ 8 L ' 8 u ^ k k ' l k

r ^ ( Fl-L\L· i F K - K \ k Y \ Y ) L' ^ \ Y ) K (2.7) where: = f ,

k

- f - ,

_{d lL}_’ and Fi = for i = K, L.

By letting a = , and

(1

— cv) = , where a and {1 — a) stand for the factor-output shares of labor cind capital respectively. Thus substituting (2.4) into (2.7), we get:

gr = oc-n + { l - a)gi. (2.8)

where the letter g stands for growth. That is the output growth will be the weighted average of both the growth in capital and labor.

On the other hand, assuming perfect competition would have the implication that payments to the factors are equal to the value of their marginal products. Equivalently, using (2.2) we can write the eqiuition in (2.9) and (2.10):

(27)

₁₈

w =

6Y

s ( f )

- m w =

K

J j f { k) - k · . f { k )

= m

. f { k ) (2.9) 6Y S( f ) , r = L . ^ ^ . f ' ( k ) SK 6K (2,10) Using equation (2.4) and (2.6), the growth in the capital-labor ratio Ccin be e.xpressed as in equation (

2

.

11

): 9k u ih. ^ ___ ____ 6 lL ___ Su k k \^k l- Lk I± L L

2

I± L gi< - 9 L s{k)S(k) k — 7 — n s{k) f{k) - {n + -f)k k

(

2

.

11

)

Now, assuming that before the tax reform, our economy was in a stable state governed by a balanced growth characterized by gy = = gi — n and <

7

r· =

0

. Using equation (2.11), this would imply that;

s{k) j(k) — (n -|- 'y)k —

0

(

2

.

12

) That is gross savings per worker are just the necessary amount to offset depreciation and the capital-deepening necessary to allow for the growth of the labor force.

(28)

₁₉

2.1.2 Extension of the Model by Imposing a Capital-tax

(¿ c )·

Until this stage, we have assumed that there are no taxes. The imposition of a tax on capital would effect the capital-labor ratio (k) and the saving function. The rationelle behind the first cirgument is that a capital tax disturbs the relative cost structure of the factors of the production, leading to a factor substitution that changes the previously prevailing capital-labor ratio (k). The fact that the savings function will be affected by the Ccipital tax can be rationalized by using a Kaldoricin approach where there is a propensity to save out of profits and s«, out of wages. This results in gross savings to have the following form:

s{k) ■ f { k) = Sr(l - a) · f { k) -f · a · f { k) (2.1.3) A capital tax tc will affect the factor output shares, thus affecting savings. Therefore , the steady-state configuration in (12) can be rewritten after the introduction of a capital tax tc as:

’[k{tc), tc] ■ f[k{tc)] - (n -b

7

) · k{tc) =

0

(2.14)

Now, let us study the effect of an increase in the capital-tax tc, on the steacl

3

^-state pcittern of growth. This is done by partially differentiating (14) with respect to U:

ds dk ds d f dk dk - • - V + s r - / + ‘ " 5 i ' 5 r - ( ’‘ - 7 ) 5 T = o dk dtc or dtc dtc (2.15) where: = |f and ft _ “ c)A; ’ - at. dk [Sk- f + S- f - (” +

7

) ] ^ = ■ / (2.16)

If we assume that an increase in tc will decrease the propensity to siwe (i.e. < 0 ), then the sign of will depend on the sign of A - Sk · f + s ■ f - {n +

7

). Note

(29)

20

that A is just the response of the steady state to a change in capital-labor ratio (differentiating (12) with respect to A: ). The stability of the stable-state would require gross savings in the case of a decrease in k, to be at a level higher than the one needed for depreciation and compensating for the labor growth, in order to allow for capital accumulation and re-establishment of the steady-state capital-labor ratio. This would imply that A is negative. Hence, an increase in the tax on capital would lead to a decrease in the capital-labor ratio.

Now, we .see how will this decrease in k initiated by an increase in tc , affect the wcige-capital return ratio, ( f ) . Using (9) and (10), (^^jcan be expressed as:

(2.17) ^ _ m - k · f i k )

r f'{k)

Differentiating (2.17) with respect to k and rearranging yields the equation i (2.18):

in

^ [ /- ( t ) - f'(k) - kr(k)\ f'{h) - [f(k) - kf ( k ) ] r [ k )

dk [ { ' { k) f

where the positive sign of (17) is implied by the concavity of / . Thus an increase in ic will decrease k , and hence decrease .

Now, for the sake of completeness, we study separately the effect of the change in the capital-labor ratio {k) on wages (¿y), the gross rate of return to capital (?’ ) cind the net after-tcix rate of return to capital (r* =

(1

= i^) r).

Using equation (2.9), the response of wages can be written as:

div

dk = r { k ) - [ k f i k ) + f { k ) ]

= - k f " { k ) > 0 (2.19)

where the positive sign in (19) follows from the concavity of J.

Using the e.xpression in (10), the effect ol a change in the capital-labor ratio can be expressed as:

(30)

CHAPTER 2. THEORETICAL EOUNDATIONS

₂₁

dv

â j = m < o (

2

.

20

)

Finally, the effect of a change in the capital-labor ration on the net after-tax rate ol return to capital can be expressed straightforward from the special form of the savings function (1.3) assuming that sf io) = 0 (a quite fair cissumption):

= ,s,..(l - a) f { k)

= S r . r * . k

Using (

2

.

12

), r*can be expressed as;

(

2

.

21

)

i.e., r

* (^ +

7

) r = --- = cte

is not affected by a change in the capital-labor ratio (k).

(2.22)

2.1.3 Results:

Throughout the last subsection, the following results were derived:

1

. ^ <

0

i.e. an increase in the tax on capital will induce a decline in the capital-labor ratio.

d( — )

2

. >

0

i.e. a decline in the capital-labor ratio will generate a decline in the wage-capital return ratio

3. I f > 0 i.e. a decline in the capital-labor ratio causes a decline in wage earnings.

4. I f < 0 i.e. a decline in the capital-labor ratio will lead to an increase in the gross return to capital.

5. ^ = 0 i.e. a change in the Ccipitcil-hibor ratio does not affect the net return to Ccipital (if we assume that the marginal propensity to save out of wages: S-w = 0).

(31)

₉₉

In other words, and to establish the link with our topic, the imposition or an increase in the Ccipital tax as a part of the tax package that usually accompanies an economic reform will be totally shifted to the labor class. As a result, in our framework, we can conclude that although the tax was imposed initially on capital, it was shifted, and the actiuil payers come out to be the wage-earners.

Thus, if we assume that this mechanism of tax shifting is not discovered by the labor class, they may vote for a certain tax reform ex-ante, but expost and after realizing that their costs are actually more than what they expected, may be willing to ’’ short-cut ’’ the reform and return to the status-quo.

2.2 Indirect Taxes (Seignorage): W ho Actually

pays for Them?

Now we turn our attention to another source of public financing, namely seignorage. In Turkey a large share of public deficit is financed through the centrcil bank. The ability of the central bank to do so stems from two sources. First, the central bank captures a fiirge inflow of foreign exchange resources due to the fact that it borrows from abroad, hold the foreign exchange deposits of Turks working abroad, and imposes the revenue requirements on foreign exchange deposits held in commercial ba,nks. Second, the central bank also finances the public sector directly through money creation.

Total seignorage as a percentage of GNP is measured by

J ’

stcinds for the base money at time ¿, including currency, required reserves, and excess reserves. In Turkey, total seignorage was high in 1979 ¿incl amounting to

4.6

percent of GNP. Such revenue then oscillated between 3.2 and 4.2 percent of GNP between 1980 and 1984 and declined steadily afterward to an estimated 1.7 percent of GNP in the first semester of 1987.This decline in the seignorage revenue can be exphiined by two factors; the revenue requirement which includes the reserve requirement ratios, computation method, ¿md the banks complicince with the rules; and the portfolio behavior of asset holders, which is influenced by inflation, interest rates, output growth, and the emergence of close substitutes for domestic money.

(32)

CHAPTER 2. THEORETICAL FOUNDATIONS 23

This latter point will be the main focus of this section.

Before proceeding further, let us stop for a moment and try to analyze the reasons behind this decline in total seignorage revenue. Total seignorage ] can be broken into two components:The inflation tax and the real growth of base money , where pt is the consumer price index (CPI) annual inflation rate. The inflation tax stems from the fact that asset holders try to preserve the real value of their money balances.The amount demanded to reach this end depends positively on the inflation rate and is supplied by the central bank at negligible cost. It can then be transferred to the public sector to hnance the purchase of goods cind services. The second component measures the variation in the real stock of money. These variations can be caused by variations in real output, inflation or interest rates, as well as financial innovations or changes in the reserve requirement regime. Real output, inflation, and interest rates affect demand for base money by affecting demand for financial assets. The reserve requirement ratio directly affects the demand for base money through changing the required reserves for a given real stock of deposits. Finally, innovations in the financial system may affect demand foi- base money either directly (for example, the development of an interbank market reduces the bank’s excess reserve for the same real stock of deposits) or indirectl}'^ (for example, better cash management may reduce the firm’s demand for sight deposits).

In Turkey the dominant component in seignorage revenue originates in the inflation tax due to the inflation rates that prevailed throughout the last one and a half decade. A genuine question that should have risen at this point is what relates the inflation tax to our topic namely reform non-sustenance? Well our claim is that the inflation tax is regressive in nature which concentrates its burden mainly on the poor class. The rich class having the advantage of access to financial adaptation, can hedge a big proportion of their income against inflation. This fact can be claimed to enhance distribution inequality which generates political instability that was shown to affect reform sustainability (Alessina and Perotti (1993)). In our analysis we bring this point further and claim that the actual reform period will be higher than that expected by the poor (wage-earners) class prior to the reform adoption. In what remains in this section we will review an article by Laban and Sturzenegger (1994) and show how their model can be modified to generate our main target in this section that the actual reform duration will be more than expected.

(33)

2.2.1 The Model:

Liibari Sturzenegger (1994) assume that there are two types of agents that they call poor (p) and rich (r). The rich differ from the poor in that they have access to a tax shielding technology which allows them to reduce optimally the base upon which a distortionary tcix is levied. This technology is assumed not available for the poor or too costly. The poor and the rich receive cin endowment Cp ¿xnd respectively at the beginning of each period. In addition to that, before stabilization is adopted, the government collects an amount g per period financed through a distortionary tax levied on both groups, which it transfers to the poor.

The issue that both types of agents have to decide on in the beginning of each period is whether to stabilize or not. Stabilization which balances the government lj)udget is achieved through a combination of tax increases and expenditui'e cuts. It is assumed that there is perfect information and the only source of uncertainty is due to the fact that agents are not sure a priori about the effectiveness of the government in enforcing the stabilization program agreements.

Given this setup, the government’s budget constraint in each period t is given by;

9 = TTi İTp A" fi)· Ft) (2.23)

1

+ 7Ti

where g is the level of transfers to the poor the government collects each period, is the tax rate which is supposed to be determined endogenously, and depends on the aggregate level of tax evasion chosen by the rich at period f , F( · tt can be interpreted as the inflation tax, g as seignorage revenue and the tax evasion technology as a process of financial adaptation that take the form of capital flight, changes in transaction technologies, dolarization or purchases of real or indexed assets.

Engaging in financial adaptation by the rich is assumed not to be costless. Using an amount in financial adaptation any given time t by a rich agent, induces a cost c(ft, kt), where kt = ^ F, is the accumulated level of financial adaptation until

z=0

the beginning of the period t. This may be understood as a process of ’’ learning by doing” or the development of a wider viiriety of financial and transaction institutions. The following cicceptable assumptions are imposed on the cost function c{ft,kt):

(34)

individuals are assumed to be identical, in equilibrium we should have ft — F\ for any t.

If stabilization is not agreed upon, the poor’s share of distortionary taxation iit time t would be 9t such that:

0t = (2.24)

Cp + Cr — Et where

(1

— Ot) is the share born by the rich.

in addition to the distortionary tax, inflation is also assumed to have distortionary effects which generate welfare losses. Those losses are incorporates as a reduction of in the endowments of each individual where >

₀

and >

₀

.

Built on those assumptions, the pre-stabilization utility flows for the representa tive agents in period t are given by:

Up(ct) = Ui ep-fixt) + g - &tg) (2.25)

Ur(ct) = U{er - f{irt) -

(1

- et)g - c{ft, kt)) (2.26) where U(.) is assumed to a twice-differentiable, monotonically increasing, constant absolute risk aversion utility function, with risk aversion coefficient

7

>

0

.

The writers assume for simplicity that post-stabilization taxes fall completely on the rich. The two groups bargain on the expected level of post-stabilization redistribution of income gU Due to the existence of instrument uncertciinty, the bargaining process will be built on expected values. Instrument uncertainty is conceptualized by assuming that ğ the amount of post stabilization transfers, equcils:

g = tCr — a (2.27)

where a is a random variable with zero conditional mean and variance (T^. Ihus, il stabilization is achieved, expected utilities in each period, conditional on the prior information available, would be lor the poor and rich respectively as follows:

(35)

EU{er + — U[er + g^) (2.29)

where g'^ cind .stand for the certainty equivalents. Thus given that ^ is the risk premium both agents are willing to pay in order to avoid uncertainty, we have that:

g; ^ ter 1^1 (2..30)

fr = + 7<^«₂ (2,31)

Laban and Sturzenegger work with a two-period time dimension cind derive lour main propositions implied by their model. Without entering in further proof technicalities, we state the propositions and discuss in turn their implications briefly:

P r o p o s itio n

1

: Provided that the assumptions on U{.) and c { f , K ) cire satisfied, if stabilization is not achieved in either period, we have for the stable monetary equilibrium that:

1

. /(* >

0

for t = 1,2 (positive equilibrium level of financial adaptation);

2

. /* > fy (increasing equilibrium level of financial adaptation);

3

.

7

T

2

>

7

Ti (inflation endogenously increases);

4. $2 > ill (regressive impact of financial adaptation);

This means that the optimal degree of financial adaptation will be strictly positive iind increasing over time. This is due to the fact that in engaging in such process, the agent will equalize costs to benefits; and since the cost is decreasing (remember the assumptions we posited on c ( /, K), use of financial adaptation technology would be increasing with cin accelerating rate of inflation.

P r o p o s itio n 2: Provided that <

2

(/!>(

7

r

2

)

4

- c ( /

2

, /T

2

) there exists a nonempty set of possible agreements to achieve stabilization in period

2

.

This proposition implies that in order to have an incentive to stabilize, the economy-wide gains from stabilization (sum of the distortions and evasion costs)

(36)

should exceed the risk premium in engaging in an uncertain reform. Thus the reform will have more chances to be passed depending on;

1

. The size of the distortion (

2

(^(

7

T

2

)).

2

. The cost of financial adaptation

3. The degree of risk civersion or instrument uncertainty (

7

<t^).

P r o p o s itio n 3: Given the results of proposition

1

, ¿

2

p < Gp

In other words, this proposition states that the greater the level of financicil adaptation, the lower the expected level of transfers the poor are willing to agree upon in order to stabilize. That is, the bargaining position of the poor progressively deteriox'ates.

P r o p o s itio n 4: Even if there exists a non-empty set of possible agreements to stabilize in period

2

, it may still be an equilibrium strategy to delay stabilization in period

1

.

Although this proposition seems counterintuitive; technically for such a case to rise all what we need is that the conditions in proposition

2

be satisfied in the second period, but not in the first.

Now, to recapitulate the míiin ideas relevant to our topic, Laban and Sturzenegger show that with the availability of financial adaptation to a certain group in the society-most likely the rich-, the poor will be forced to accept less favorable outcomes, in order to escape the deterioration costs of the ever-increasing inflation.

2.2.2 Extension of the Model:

Let’s now turn to our target and try to show how the actual reform duration may exceed that expected. For this, let’s assume that as a part of the reform financing program, the government intends to collect a certain share of the reform costs through seignorage or the inflation tax. If we assume that there is no access to financial adaptation by neither of the two groups then seignorage collection at eacli time period is:

(37)

St — TTi

1

+ (Cp +

6

r) (2.32) Under such circumstances, and if the inflation rate is expected to be held at a constant level, the reform expected duration will be just that period, n®, sufficient to collect the necessary revenues S:

rp = S

■ sti= i+TTt7Tt (ep + Cr))

(2.33)

Now we introduce financial adaptation to our analysis. In assuming the absence of such a process in the pre-reform setup we were in fact making a quite implausible cissurnption. However, we assume here that engaging in financial adaptation is not costless, and that the cost function has three cornponents-in contrast to the component cost function introduced by Laban and Sturzenegger-. First, the cost will depend on the amount that is to be evaded, /¡. The second component is a stock variable which refers to the knowledge and ‘know-how” accumulated

t-i through previous evasion experiences, and hence can be expressed as F^.

;=:0

Finall}^ the third component is a reform-specific component which captures the new adaptation alternatives that the reform environment induces. This may be ‘ dolarization” if the reform program contains as one of its parts the liberalization of foreign exchange markets. This third component is called K ‘^ and is considered

¿IS cl ‘ cumulative” process that can be expressed as = Y , where is the time

z=to

period where the reform starts. Thus the cost of engaging in financial adaptation can be written as:

ct = c { f t , K l , D , - K f ) (2.34) where Dt is a dummy variable that takes a value of

1

during the reform period.

Now, a critical step in our analysis is in the choice of the assumptions that we make about the behavior of the cost function. Due to the way we have characterized Cl, the following assumptions seem to be intuitively ’genuine’ :

The higher the amount to be evaded, the higher will be the cost: Cj > 0. Moreover the higher is the amount to be evaded, the higher will be the marginal

(38)

cost of evading an extra dollar of money: C // >

0

.

• Prior to the reform, the more individual engages in such a process, the more experience and know-how he will gain, and hence the lower will be his costs: C'ki < 0. However, as time passes, the amount of information that can be gained by experiencing tends to be diminishing and hence: >

0

.

• After the reform and with the emergence of new financial adaptation methods, the cost of evasion will be decreasing as people get acquainted with them: Cf,2 <

₀

.

From the assumption that Ckiki > 0, it follows that if the society is acquainted with financial adaptation a long time before the reform is proposed, then the level of evasion in the few years preceding the reform tends to be rather constant: Cki — 0. Thus if we assume that people ignore that the reform will affect the costs of evasion, which is the only source of imperfect information in our analysis; then the assumption that we have made in (32) about seignorage collection will not be so ‘ disastrous” . In fact, the expected reform duration would be:

5 n

{ l f ^ - ( e p + e r - F ) } (2.35) where E' is the pre-reform, nearly constant level of financial adaptation.

Now after the reform is adopted, the level of evasion will be increasing, but in order to sustain the planned level of seignorage revenues the inflation rate will be increasing either endogenously or through government monetary actions, with the burden of the inflation being more and more regressive. The planned level of seignorage revenues will be sustainable if we assume that the inflation rate can increase or be increased infinitely. However it is frequently the case that there is a maximum socially bearable inflation level if, beyond which it is politiccilly and socially non-feasible to increase the inflation level. Thus faced with the increasing- level of financial adaptation and a constraint on the level of inflation, the government may fail to collect the pre-reform planned seignorage income per-period, sC For this to take place cdl what we need is that at a specific period during the reform we have:

Why are reforms non-sustained?: a case study of Turkey

ЩУ¥· I p “ ^ r p ñ p f j f ; ІІ^Щ «

*. â fíi^ü

WHY ARE REFORMS NON-SUSTAINED?; A CASE

STUDY OF TURKEY

E

S

S

IN

t l j ^

ABSTRACT

ÖZET

ACKNOWLEDGEMENT

Contents

1.1

2

1

1.1

2

1

1.2

1.2

1

2.1

7

10

11

2.1

2

1.1

2

1.2

21

2.2

22

3.2

4.2

Chapter 1

Introduction: Literature Review

1.1

structural Features:

1.1.1

Pre-reform Economic Environment:

2

4

1.1.2

Institutional Structure:

5

1.1.3

Nature of the Political System:

1.2

Economic Policy Choices:

1.2.1

Speed, Depth and Sequencing of reform:

1.2.2

Distributional Aspects and Foreign Assistance:

1.3

The Interplay between Structural Features

and Economic Policy Choices:

1.3.1

Initial Macoeconomic Disequilibrium:

9

1.3.2

Sequencing and Speed of Reforms:

10

1.3.3

Speed of Reform in the Case of Restructuring:

11

1.3.4

Credibility and Political Uncertainty:

1.3.5

The Supply Response to Reforms:

12

1.3.6

Distributional Conflict:

1

2

3

13

1

₁₀

₁₁

₁₂

₁₃

₁₆

_(2.2)

₁₇

₁₈

₁₉

₂₁